Up to 80% of breast, bowel and prostate cancers are attributed to dietary practices, and international comparisons show strong positive associations with meat consumption. Estimates of relative risk obtained from cohort investigations are in the same direction, although generally weak, and red and processed meats rather than white meat seem to be associated with elevated risk of colon cancer. In breast cancer, there are consistent associations with total meat intake and there is evidence of a dose response. Despite these associations with meat, existing studies suggest that vegetarians do not have reduced risk of breast, bowel or prostate cancer, but there are no quantitative estimates of amounts of meat consumed by meat eaters in these cohort studies. Possible mechanisms underlying epidemiological associations include the formation of heterocyclic amines in meat when it is cooked. These heterocyclic amines require acetylation by P450 enzymes, and individuals with the fast-acetylating genotype who eat high amounts of meat may be at increased risk of large-bowel cancer. NH3 and N-nitroso compounds (NOC) formed from residues by bacteria in the large bowel and probably also important. NH3 is a promotor of large-bowel tumours chemically induced by NOC, and some of the chromosomal mutations found in human colo-rectal cancer are consistent with effects of NOC and heterocyclic amines. However, the type, amount, and cooking method of meat or protein associated with increased risk are not certain. The effects of high levels of meat on NH3 and NOC output are not reduced by increasing the amount of fermentable carbohydrate in the diet, but interaction between meat, NSP and vegetable intakes on the risk of cancer has not been studied comprehensively. The interaction between dietary low-penetrance genetic polymorphic and somatic mutation factors has also been investigated to a limited extent. Current Department of Health (1998) recommendations are that meat consumption should not rise, and that consumers at the top end of the distribution should consider a reduction in intakes.